Category Learning Selectively Enhances Representations of Boundary-Adjacent Exemplars in Early Visual Cortex.

Area of Science:

• Cognitive Neuroscience
• Computational Vision

Background:

• Category learning and visual perception are interactive, with learning improving discrimination of relevant stimulus features.
• Perceptual enhancements are linked to visual system plasticity, but underlying mechanisms remain unclear.

Purpose of the Study:

• To investigate if category learning sharpens sensory representations in the visual cortex.
• To determine if this sharpening is specific to the learned category rule and occurs during active learning.

Main Methods:

• Used functional magnetic resonance imaging (fMRI) and a computational vision model.
• Quantified changes in orientation representations in visual cortex (V1-V3).
• Subjects learned to categorize stimuli using either an orientation or spatial frequency rule.

Main Results:

• Orientation representations were enhanced in visual cortex for subjects learning the orientation rule.
• Representations were sharper for stimuli near the category boundary compared to those at category centers.
• This sharpening effect was specific to the orientation rule learners.

Conclusions:

• Category learning induces representational sharpening in visual cortex, particularly for behaviorally relevant stimuli.
• This plasticity efficiently enhances sensory populations tuned to critical regions of feature space.
• The findings support a model where learning selectively modulates sensory processing based on task demands.